svaante / decision-tree-id3 Goto Github PK

I installed in a virtual enviroment. But I am unable to use the module. Please help

HI, I think the 134th line in the function __prune (in tree.py), the condition for pruning should be changed from 'node_error_rate < children_error_rate' to 'node_error_rate<=children_error_rate'. Because node_error_rate==children_error_rate means that increase the complexity of the model won't increase the accuracy of classification. Thus, to prevent from overfittting, the condition should be changed to 'node_error_rate<=children_error_rate'.

Was using this library and I noticed that there aren't parameters in the documentation for Id3Estimator(). Are there any parameters that we can set? Maybe something similar to sklearn's DecisionTreeClassifier() where you can choose max_depth, criterion, etc.

Hello,

By importing id3 I encounter this error on my setup. Is it possible that importing six from sklearn.externals is outdated?

File "/usr/local/lib/python3.7/site-packages/id3/__init__.py", line 3, in <module> from .export import export_graphviz File "/usr/local/lib/python3.7/site-packages/id3/export.py", line 1, in <module> from sklearn.externals import six ImportError: cannot import name 'six' from 'sklearn.externals' (/usr/local/lib/python3.7/site-packages/sklearn/externals/__init__.py)

thank you

When I run clf = Id3Estimator(is_repeating=True) and clf = Id3Estimator(is_repeating=False) with the toy data in the demos, I get same decision trees which both contain repeating features.

I got problem when split dataset into train and validation set. If validation set contains attribute value that never seen in train set, this exception will occurs. This problem occurs because LabelEncoder execute during training (fit). How to make sure all attribute value from my dataset converted by LabelEncoder ?